Outcome of Anorexia Nervosa: results of a 5 year follow-up study

Transcription

1 Outcome of Anorexia Nervosa: results of a 5 year follow-up study Annemarie A. van Elburg, Jacquelien J.G. Hillebrand, Karlijn Kampman, Marinus J.C. Eijkemans, Martien J.H. Kas, Herman van Engeland, Hans W. Hoek. (submitted for publication) 8

2 Chapter 8 Abstract We describe the findings of a 5-year follow-up study of severe anorexia nervosa (AN) patients referred to two specialized treatment centers in the Netherlands. Our aim was to evaluate their current state of illness and to identify potential predictors of recovery. We contacted patients who participated in a cohort study during the first year after admission. Body weight and return of menses, as well as psychological features such as body attitude, psychopathology and mood states were investigated at 5 year followup. All 61original patients were contacted: 59% of the patients were weight and cycle recovered, 9.8% was only weight recovered, and 31.1% was not recovered. None of the original participants had died. Of the 61 original patients, 54 cooperated in a further examination of their clinical and psychological states. 16 respondents still met AN criteria, 12 had an eating disorder not otherwhise specified and none of the respondents suffered from bulimia nervosa. Despite the high percentage of weight recovery, many respondents manifested ppsychopathology and a severely disturbed body attitude. Respondents also frequently reported suffering from depression, anxiety, obsessivecompulsive and personality disorders, and feeling more tired and tense than years before. We found no predictors of treatment outcome other than relapse, but discovered that patients whose outcome changed from weight and cycle recovered after 1 year into not recovered after 5 years had been rated as more hyperactive at the 1 year time point. In conclusion this study shows that five years after admission, many AN patients recovered in body weight and menses. The majority of the participants however still suffer from severe psychopathology. The 5-year outcome could not be predicted by treatment outcome after 1 year, co-morbidity, or duration of illness. To predict treatment outcome and promote evidence-based treatment, more long-term follow-up studies of outcome in AN patients, with particular focus on the somatic as well as psychological parameters of the disease, are needed. 150

3 Outcome of Anorexia Nervosa: results of a 5 year follow-up study Introduction Recovery from AN in general takes a long time; it has been reported that stable physical recovery is reached after on average 4.7 years and psychosocial recovery after 6.6 years (Strober et al., 1997; Fennig et al., 2002; Eckert et al., 1995). Final outcome figures of AN leave room for improvement. According to Steinhausen (Steinhausen 2002), 20.8% (0-79%) of AN patients remain chronically ill and 5.3% (0-22%) die as a consequence of starvation or suicide (Birmingham et al., 2005; Steinhausen 2002). Chances for recovery range from 0-92%, averaging at 46.5%. Recovery data in younger patient groups are more optimistic, reaching to 60% chance for complete recovery (Steinhausen 1997; Steinhausen 2002). Full recovery of AN is in general defined as recovery of normal body weight for age and return of regular menstruation for at least 6 months (Morgan & Russell 1975). though normalization of anorexic cognitions and eating patterns are also important for returning to normal life (Pike 1998). Despite weight recovery, anorexic cognitions and abnormal eating patterns are still frequently observed in somatically recovered patients. Besides cognitions and eating patterns, a number of authors emphasize the importance of the patient s social environment in relation to recovery (Noordenbos & Seubring 2006; Norring & Sohlberg 1993; Steinhausen & Glanville 1983a; Steinhausen & Glanville 1983b). Furthermore a significant comorbidity of anxiety and affective disorders with AN is frequently noted (Halmi et al., 1991; Herpertz-Dahlmann et al., 2001; Kaye et al., 2004; Keel et al., 2005). A meaningful construct of recovery needs to consider not only the severity of the symptoms but also the duration of improvement. Many studies indeed report that AN patients are at substantial risk of relapse. On average 35 50% of patients experience one or more relapses, depending on the follow-up time and definition of relapse (Keel et al., 2005; Norring & Sohlberg 1993; Pike 1998; Strober et al., 1997). Besides relapses into AN, a crossover to full blown bulimia nervosa (BN) is also frequently observed. Tozzi et al. described that 36% of their restrictive AN sample (ANr) developed BN, the majority within the first 5 years of the illness (Tozzi et al., 2005). The duration of a symptom-free state appears to be an important component of a definition of recovery. However, several of the studies that incorporate duration of symptom- 151

4 Chapter 8 free state into the assessment process define recovery as maintaining a symptom-free state for only eight consecutive weeks (Strober et al., 1997). Although this definition might serve inter-study comparisons, it seems inappropriate to label patients as recovered from AN after such a short symptom-free period. Indeed, many argue that meaningful evaluations of recovery from eating disorders can be made only after long-term (several years) follow-up (Kreipe et al., 1989; Pike 1998). Therefore, it is necessary to clarify the use of the term recovery in outcome research and initiate more studies assessing change in outcome over a series of short follow-up intervals (Herzog et al., 1988). There are specific features of the illness that seem to influence outcome. Evidence exists that an early onset and early treatment of AN lead to a better prognosis of treatment outcome, as do a lack of symptomal binging or hyperactivity (Steinhausen 2002; Strober et al., 1997; Wentz et al., 2001). The present study was performed to investigate the course and outcome of AN in 61 former adolescent patients approximately 5 years following admission to our tertiary specilized treatment centres. We examined current body weight and menses as well as current psychopathology and investigated the presence of predictors of outcome. Materials and methods Participants The study sample consists of 61 female AN patients who participated in a cohort study as described elsewhere (van Elburg et al., 2007). This original study was conducted between February 2000 and January 2003 and focused on psychoneuroendocrinological changes during treatment of AN in two specialized treatment centers, one for adolescents years (UMCU), 50.8% (n=31), and one for patients older than 17 years (Rintveld), 49.2% (n=30). Approximately five years after admission (T5) all participants were traced back and asked to participate in a follow-up study to investigate their current state of illness and psychological functioning. 152

5 Outcome of Anorexia Nervosa: results of a 5 year follow-up study At admission (=T0) all 61 participants fulfilled DSM-IV criteria of AN. Of the total sample, 68.9% (n=42) was diagnosed with restricting type AN, whereas the remaining 31.1% (n=19) was diagnosed with AN of the binge/purging type. Average age of admission was 18.2 ± 3.1 years, corresponding BMI was on average 15.4 ± 1.3 kg/m 2 (range ), and average illness duration (in terms of duration of amenorrhea prior to admission) was 24 ± 22 months. Every participant was a Dutch citizen; only 2 were of non-caucasian descent. Patients with physical or psychiatric co-morbidities other than anxiety or depression as well as male patients were excluded from joining the study at T0. The longitudinal cohort study lasted until the moment that body weight and menses were recovered (T1), with a maximum duration of one year. It was approved by the Dutch Medical Ethics Committee. All participants (and their parents in the case of minors) gave their informed consent and were informed about the possibility of a future follow-up study. All 61 participants received clinical treatment conducted by two multidisciplinary and specialized eating disorder teams. Both teams took an integrated approach aimed at recovery of a normal body weight, eating pattern, and body attitude, as well as normalizing family relations and further development of social skills at an inpatient and/or outpatient level. Follow-up measurement Approximately 5 years after admission, all 61 participants of the longitudinal cohort study were informed about a follow-up study by an announcement letter mailed to their last known address. This letter gave an address which participants could use if they did not want to be further contacted. Thereafter participants were contacted by phone for further explanation of the study and were asked to participate. A set of (self-report) questionnaires was mailed to participants who agreed to take part to investigate the current state and course of their AN. If forms were not returned within 3 weeks, participants were contacted again. Participants who did not wish to take part in the follow-up were asked for their reasons as well as their current state of illness. 153

6 Chapter 8 The test battery consisted of several items to assess Morgan Russell criteria for body weight recovery and regular menstruation (Morgan & Russell 1975). Good outcome was defined as cyclic menstruation and a minimal BMI of 18.5 (T5) or BMI computed into z-scores (T1) (van Buuren & Fredriks 2001) (WCR: weight and cycle recovered). Intermediate outcome was defined as a minimal BMI of 18.5 without (regular) menstruation (WR: weight recovered). Poor outcome was defined as BMI below 18.5 and absence of menstruation (NR: no recovery). When respondents gave no specific information about their body weight, an estimation of BMI was made by the principal researcher (AvE), based either available information from the records or on visual examination of the respondent at the reunion of study participants a few months later. Body weight data from these respondents were used to calculate BMI categories only. All respondents were asked about their use of oral contraception. If respondents used oral contraception and reported menses, their BMI values were first examined to judge whether a natural menses would be possible. Secondly, their BMI and menstrual state at T1 (when contraception was prohibited) were compared with their BMI at T5. Once a respondent had regular menses at T1 and BMI at T5 was consistent with BMI at T1, a natural menses was assumed. Respondents were asked about current treatment and the existence of other eating disorders or other psychiatric problems. Eating behavior, anorexic cognitions, body attitude (Dutch version of Body Attitude Test, BAT) (Probst 1995), happiness (subjective grades: 0= very unhappy, 10=very happy), social and occupational adjustment, as well as general psychopathology (Dutch version of Symptom Checklist-90, SCL90, (Arrindell 2005) and current mood state (Dutch version of Profile of Mood State, POMS, (van der Ark LA 1995) were investigated. The POMS data were scored on a visual analogue scale and converted to stanines to obtain high and low scores. In this paper only raw scores are presented (Little & Penman 1989). We also asked whether respondents had experienced a relapse (defined as a body weight below BMI 18.5 after achieving a significant initial response, and/or as amenorrhea) in the period up to T5. Finally we investigated whether participants (log) leptin levels and nurse ratings of their hyperactivity at T1 (Van Elburg AA 2007) were predictors of 5 year outcome. 154

7 Outcome of Anorexia Nervosa: results of a 5 year follow-up study Statistics The participants were divided into three categories of recovery. In this paper, data will be presented for all three categories, as well as for the two weight recovered categories combined (when there was no significant difference between WCR and WR). The program SPSS 14.0 was used for statistical analysis. Normal distribution of all data was investigated. Descriptive analyses, ANOVA, t-tests for paired values or independent values or non-parametric Mann-Whitney U statistics were performed. For correlation analysis Pearson coefficients and chi-square statistics were evaluated. Stepwise linear regression was performed to investigate putative predictors of outcome. A p value of <0.05 was considered significant. Results Traceability: We collected information on body weight and menses recovery of all 61 former patients (100% traceability) on average 4.8 ± 0.8 years after admission. A further investigation of psychological features was performed in 88.5% (n=54) of the participants ( respondents ), at T5. The remaining 11.5% (n=7) of participants ( rejectors ) did not want to take part further for several reasons (they were still suffering severely from AN (n=2), were currently healthy and did not want to be remembered of AN (n=4), or for unknown reasons (n=1)). The rejectors did not differ from the respondents in type of AN, age, BMI at T0, nor in outcome at T1 or T5 (data not shown). Criteria of outcome: BMI and menses At admission to the treatment centers (T0), participants fulfilled DSM-IV criteria for AN. At the end of the cohort study (T1), on average 33 weeks later, 39.3% (n=24) of participants were recovered in body weight and menstruation (WCR), 29.5% (n=18) recovered in body weight but still showed amenorrhea or menstrual irregularity (WR), and 31.1% (n=19) had 155

8 Chapter 8 not recovered in body weight or menstruation (NR) (Figure 1, (van Elburg et al., 2007)). At five years follow-up (T5), 59.0% (n=36) of them had a normal body weight and menstruation (WCR), 9.8% (n=6) had a normal body weight but showed lack of menstruation or menstrual irregularity (WR), and 31.1% (n=19) had a subnormal body weight or menstruation (Figure 1). Hence, outcome at T1 was not stable. Figure 1 depicts the shift of patients among the three outcome categories; it shows that 42% of patients who were labeled NR at T1 developed into WCR after treatment.. However, 26% of the patients labeled WCR at T1 have a poor outcome at T5. Interestingly, it appeared that these patients were rated as more hyperactive at T1 compared to patients with persistent WCR outcome (t(19)=2.894, p<0.009), whereas they did not differ in any other parameter. Average BMI at T5 was 18.6 ±3.2, which was significantly higher than BMI at T0 but not BMI at T1 (z=-4.929, p<0.001, z=-0.175, n.s.). Average BMI of the WCR and WR group at T5 was 20.2±2.3, versus 15.5±2.1 of the NR group (z=-5.546, p<0.001). At T5 30.6% (n=15) of 54 respondents still showed an extreme underweight (BMI<17.5) (Table 1). Outcome at T5 (BMI and menses) was not related to the subtype of AN at T0, age at T0, the original treatment centre, duration of illness, or co-morbidity at T0 (data not shown). BMI at T5 was related to BMI at T1 (r=0.290 p=0.04), although the association was rather weak. In total, 59% of the respondents had a natural menses at T5. As of its definition, all members of the WCR group showed regular menses (although some used oral contraceptives). Seven other participants reported menses and oral contraceptives use, however, the existence of natural menses in these participants was highly unlikely because of their low BMI. (BMI<17.5). In one respondent, natural menses could not be ascertained despite weight recovery. 156

9 Outcome of Anorexia Nervosa: results of a 5 year follow-up study Figure 1. Outcome groups of 61 patients with AN, at start (T0) after one year (T1) and 5 years. (T5). WCR: weight and cycle recovery, WR: weight recovery, amenorrhea, NR: no weight recovery, amenorrhea. Course of the illness: relapse, further treatment At T5 16 (of 61) participants still fulfilled DSM-IV criteria for AN (13 ANR, 3 ANP). One of these participants crossed over from ANP into ANR. No respondents showed BN (although a history of BN was observed in one respondent) and 12 respondents showed characteristics of eating disorders not otherwhise specified (EDNOS). The majority of the 54 respondents (72.2% (n=39) had experienced one or more relapses following remission. The experience of a relapse did not correlate with the diagnosis at T0, the outcome at T1, age, or treatment centre (data not shown), but did correlate with the outcome at T5 (χ 2 =35.507, p<0.001). Respondents from the WCR and WR group experienced fewer relapses than the respondents from the NR group. The majority of the respondents (72.2% (n=39) continued treatment for AN elsewhere in the period up to T5, ranging from individual therapy (n=17) to group therapy (n=2) and to inpatient treatment (n=8), or a combination of these (n=14). At T5, 33.3% (n=18) of the respondents were still in treatment for an eating disorder. Respondents suffered mainly from depression (33.3%, n=18), anxiety disorders other than OCD (18.5%, n=10), OCD (11.1%, n=6) and personality disorders (9.3%, n=5). Hence, 157

10 Chapter % (n=13) of the respondents started treatment for co-morbid problems after T1 and at T5 18.5% (n=10) still used treatment, ranging from low-frequency individual outpatient treatment to inpatient treatment. The frequency of current (self-reported) co-morbid problems did not differ between WCR, WR and NR groups. In addition, 37.0% (n=20) of the respondents used medication at T5, mainly selective serotonin reuptake inhibitors (SSRIs) (n=10), tricyclic antidepressants (TCAs) (n=6) and/or atypical antipsychotics (n=4). When evaluating subjective state of the disease, 24.1% (n=13) of the respondents reported that they recovered from their disease and the remaining respondents reported that the state of the disease improved (44.4%, n=24), remained unchanged (18.5%, n=10) or worsened (13%, n=7) at T5. These subjective measurements of illness state correlated with outcome at T5 (χ 2 =65.366, p<0.001), thus respondents who claimed to be recovered were mainly WCR respondents. When asked whether their current body weight was of concern, 20.4% (n=11) answered that it was still continuously of concern whereas 40.7% (n=22) of the respondents answered that it was never of concern. Furthermore, 25.9% (n=14) of the respondents answered that they never ate less than they actually should, whereas 5.6% (n=3) answered that they always ate less then they should which correlated with their outcome at T5 (χ 2 =97.582, p<0.001). Interestingly, 11.9% of the W(C)R respondents reported that their body weight was continuously of concern and 62.2% (n=23) of the WCR and WR group still restricted their diet, ranging from less than half of the time (37.8%, n=14) to more than half of the time (5.4%, n=2). 158

11 Outcome of Anorexia Nervosa: results of a 5 year follow-up study Table I. T0 T1 T5 T5 Participants 100% (n=61) 100% (n=61) 100% (n=61 α ) 88.5% (n=54 β ) Age (SD) 18.2 (3.1) 18.9 (3.1) 23.1 (3.0) 23.0 (3.0) Group WCR % (n=24) 59% (n=36) 57.4% (n=31) WR % (n=18) 9.8% (n=6) 11.1% (n=6) NR 100% (n=61) 31.1 (n=19) 31.1% (n=19) 31.5% (n=17) BMI 15.4 (1.3) 18.5 (1.8) 18.6 (3.2) χ < % (n=58) 26.2% (n=16) 31.1% (n=19) 30.6% (n=15) % (n=3) 14.8% (n=9) 10.2% (n=5) % (n=0) 34.4% (n=21) 68.8% (n=42) 28.6% (n=14) % (n=0) 24.6% (n=15) 36.7% (n=18) >25 0% (n=0) 0 % (n=0) 4.1 % (n=2) Diagnosis ANR 68.9% (n=42) 24.6% (n=15) 24.1% (n=13) ANP 31.1% (n=19) 6.6% (n=4) 5.6% (n=3) EDNOS 0 (n=0) 29.5% (n=18) 22.2% (n=12) BN 0 (n=0) 0% (n=0) 0% (n=0) Amenorrhea 100% (n=61) 44.3% (n=27) 41.0% (n=25) 42.6% (n=23) Overview of participant characteristics at admission (T0), at the end of the longitudinal cohort study (T1), and at follow-up (T5). T5 data are based on self-report. Not all participants at T5 provided BMI data. WCR: weight and cycle recovered, WR: weight recovered, NR: not recovered. α : total sample of participants which were contacted at T5, β : total sample of respondents taking part in extensive followup, χ : n=49, because exact BMI of 5 participants was not provided. ANR= anorexia nervosa restricting type; ANP= anorexia nervosa purging type; EDNOS= eating disorder Not Otherwise Specified; BN= bulimia nervosa. Considering social adjustment, 57.4% (n=31) of the respondents reported being in a stable relationship; of these 9.3% (n=5) was married, 18.5% (n=10) was living with their partner or in separate households (29.6%, n=16), and 42.6% (n=23) reported being single. One of the respondents had given birth to a child, two more were pregnant. Furthermore, 57.4% (n=31) of the respondents had a fulltime job (or went to school), 11.1% (n=6) worked more 159

12 Chapter 8 than 50%, and 14.8% (n=8) had no current occupation. Interestingly, social adjustment and occupational state did not differ between the W(C)R and NR group (z=-1.373, n.s., z= n.s.). Psychopathology, anorexic cognitions, and mood Despite recovery of body weight (and menses) in the majority of the respondents, most respondents (including WCR) still showed significantly higher levels of (subscales of) psychopathology than norm groups of healthy (female) controls. Almost half of the respondents (48 %, n=26) score extremely high on the total severity index of the SCL90. The total severity score (z= p=0.005) and subscales depression (DEP), somatic complaints (SOM), insufficiency in thinking and acting (IN), and suspicion and interpersonal sensitivity (SEN) differed significantly (p<0.05) between the W(C)R and NR group, as shown in Table II. Table II SCL90 scores of 54 respondents at 5y follow-up SCL90 at T5 Total score (n=54) WCR (n=31) WR (n=6) NR (n=17) W(C)R (n=37) Norm group Mean (sd) Mean (sd) Mean (sd) Mean (sd) Mean (sd) Mean (sd) Total (60.7) (53.4) (50.3) (61.7) (52.7) * (36.4) AGO 11.4 (5.4) 11.2 (5.3) 10.1 (3.4) 12.2 (6.2) 11.0 (5.0) 8.7 (3.4) ANX 21.8 (8.3) 20.7 (7.5) 18.1 (8.3) 25.4 (8.7) 20.2 (7.6) * 14.6 (5.7) DEP 40.5 (15.9) 36.4 (14.9) 32.0 (10.5) 51.1 (14.7) 35.6 (14.1) * 23.8 (8.6) SOM 24.2 (7.8) 22.6 (5.7) 19.6 (9.1) 28.8 (8.7) 22.0 (6.5) * 18.7 (7.1) IN 20.4 (6.5) 18.8 (5.9) 18.9 (7.0) 23.9 (6.3) 18.8 (6.0) * 14.1 (5.1) SEN 39.7 (13.4) 36.3 (12.9) 33.7 (7.6) 48.2 (12.7) 35.8 (12.0) * 26.3 (8.8) HOS 9.5 (3.5) 8.8 (2.3) 8.7 (4.6) 11.1 (4.3) 8.8 (2.8) * 7.6 (2.4) SLE 6.9 (3.5) 7.2 (3.5) 3.9 (1.9) 7.8 (3.6) 6.5 (3.5) 5.2 (2.8) AGO: agoraphobia, ANX: anxiety, DEP: depression, SOM: somatic complaints, IN: Insufficiency in thinking and acting, SEN: Suspicion and interpersonal sensitivity, HOS: hostility and SLE: sleeping problems. Norm group: healthy women who were part of a mixed norm group with average age of 43 years (range: 18-83) (Arrindell) * = sign diff from NR, p<

13 Outcome of Anorexia Nervosa: results of a 5 year follow-up study Body attitude of the respondents was investigated using the BAT, and compared with a norm group of female students. Results are depicted in Table III. The majority of the respondents (70.4%, n=38) scored in the clinical range of negative body experience. The total BAT score and BAT2 score was significantly lower in the W(C)R group compared to the NR group (t(52)=2.532, p=0.001, t(52)=4.177, p<0.001). Average values of the W(C)R group were still significantly higher than values from the healthy norm group and not different from acutely ill AN patients. Moreover 65.7% of the W(C)R participants still scored in the clinical range of a disturbed body attitude. The BAT-2 score was significantly related to BMI at T5 (r=-.335 p=0.019), was increased in participants who underwent further treatment for their eating disorder (t(52)= , p=0.007), and also related to several psychopathological subscales from SCL90 (data not shown). Table III BAT scores of respondents at 5y follow-up BAT at T5 Total (n=54) WCR (n=31) WR (n=6) NR (n=17) W(C)R Norm group Norm group (n=37) Norm 1 Norm 2 Mean (sd) Mean (sd) Mean (sd) Mean (sd) Mean (sd) Mean (sd) Mean (sd) BAT Total 50.4 (21.3) 46.5 (21.2) * 42.3 (17.5) 60.8 (20.2) 45.7 (20.4) 27.6 (14.7) 42.4 (18.5) BAT (8.6) 16.1 (8.9) 12.0 (7.7) 18.4 (8.3) 15.4 (8.7) 8.4 (6.9) 8.9 (8.9) BAT (8.3) 13.0 (7.4) 12.7 (6.2) 21.8 (7.6) 12.9 (7.1) 5.8 (3.9) 4.5 (4.5) BAT (4.8) 10.4 (4.9) 10.1 (4.5) 13.1 (4.6) 10.4 (7.4) 6.7 (2.8) 11.2 (5.1) BAT1: negative appreciation of body size, BAT2: lack of familiarity with one s own body, BAT3: general body dissatisfaction. Norm group 1: healthy women with average age of 18 (range: 12-35), Norm group 2: ANR patients who were part of a larger group of mixed eating disorders patients with average age of 24 (range: 12-35) years and BMI 17.2 (4.3) * = sign diff from NR, p<0.05 We also investigated happiness and mood states. On a grade scale from 0 to 10, respondents scored on average a `6` for happiness, with a range from 0 to 9. Respondents of the W(C)R group ranked themselves on average higher ( 6.8 ) than NR respondents ( 4.7 ) (z=-3.328, p<0.001). 161

14 Chapter 8 Next, mood was investigated using the POMS, which was also used at T0 and T1 in 51 participants. Repeated measurements analysis on three POMS time samples (T0, T1, T5) of 51 participants showed that all of the subscales of the POMS except anger changed significantly over time (see Table IV). Interestingly, Table IV also shows that average values at T5 are sometimes not different or worse from the POMS data collected at T0. Paired t- tests showed that participants at T5 report significantly more vigor (t(49)=11.431) and less depression (t(49)=-2.085), but also more tension (t(43)=3.403) and fatigue (t(49)=4.832) compared to T0 (p<0.05). Whereas respondents at T0 did not differ in depression, anger, fatigue, vigor, or tension, W(C)R participants were less depressed (t(52)=3.982 p<0.001), less angry (t(46)=4.182 p<0.001), less tired (t(52)= p=0.049), more tense (t(46)=2.549 p<0.014), and tended to experience more vigor compared to the NR participants at T5. Table IV POMS T0 (n=51) T1 (n=51) T5 (n=51)# Mean (sd) Mean (sd) Mean (sd) Depression 55.3 (15.4) 47.7 (19.0) 46.8 (27.6) * Anger 35.4 (13.5) 29.2 (13.9) 35.4 (20.6) Fatigue 28.3 (14.3) 24.6 (13.3) 44.8 (25.2) * Vigor/Activity 24.0 (7.8) 27.8 (9.5) 56.0 (19.9) * Tension 42.2 (12.0) 38.7 (12.6) 53.7 (21.4) * POMS at T5 WCR (n=31) WR (n=6) NR (n=17) W(C)R (n=37) Mean (sd) Mean (sd) Mean (sd) Mean (sd) Depression 37.0 (24.2) 35.3 (21.8) 64.9 (25.5) 36.8 (23.5) * Anger 27.5 (17.6) 28.8 (17.8) 49.6 (18.9) 27.7 (17.4) * Fatigue 40.0 (24.5) 38.0 (29.9) 54.0 (22.0) 39.7 (25.0) * Vigor/Activity 59.7 (19.3) 57.2 (25.8) 48.8 (16.5) 59.3 (20.1) Tension 48.6 (21.5) 43.9 (20.1) 63.1 (18.8) 47.8 (21.1) * # POMS data is shown of 51 participants instead of 54 participants as POMS data from 4 participants were missing at T0 and T1. * W(C)R sign different from NR, p<0.05 Predictors A stepwise multiple regression analysis was used to investigate the existence of predictors of outcome of disease at T5. Except for the experience of a relapse in the period prior to 162

15 Outcome of Anorexia Nervosa: results of a 5 year follow-up study T5, we found no predictors of outcome at T5. Thus neither duration of illness, subtype of AN, BMI at T0, age at T0, comorbidity at T0, treatment center, outcome at T1, nor BMI at T1 predicted outcome of disease at T5. Furthermore, POMS scores at T0 or T1 did not predict treatment outcome at T5 and neither did (log) plasma leptin levels and nurse ratings of patients hyperactivity at T1 Discussion In this paper we describe the results of a 5 year prospective follow-up study on adolescent AN. Five years after admission for severe AN the majority of participants had a good outcome, concerning body weight and menses, but still a large part of them suffered severely from anorexic cognitions, anxiety and depression. The strength of this study is that we managed to contact all 61 participants that were extensively studied five years before (van Elburg et al., 2007). Information on state of illness was obtained for all participants, none of the participants died, and 88.5% of the participants agreed to take part in a more extensive followup to investigate psychological features. At follow-up (T5), 59% of the participant recovered in body weight and menses (WCR), 9.8% showed a recovered body weight, but no menses (WR), and 31.1% still suffered from AN (NR). The recovery rate of our sample is comparable with the 5 yr follow-up data of Steinhausen (Steinhausen & Seidel 1993). Both studies focussed on young AN patients, which might explain the good (physical) outcome numbers. Patients in our study were on average 18.2 years old when they entered treatment at our units. It has been described previously that young AN patients (10-20 years) in general have a better outcome than older AN patients (Herpertz-Dahlmann et al., 1996; Steinhausen 1997; Steinhausen 2002; Strober et al., 1997). We found that the majority of the respondents did not meet the diagnosis of an eating disorder at T5, 16 respondents however still fulfilled the criteria for AN, and 12 respondents still showed features of an eating disorder not otherwhise specified (EDNOS) (including the WR category). A surprisingly low number (n=1) of participants developed BN after first diagnosis of AN. This is in contrast with earlier reports (Bulik et al., 1997; Keel et al., 2005; Steinhausen 163

16 Chapter 8 & Seidel 1993; Wentz et al., 2001), but in agreement with the 5 year follow-up data by Ben- Tovim and colleagues, who found only 5% of crossover between diagnoses (Ben Tovim et al., 2001). Crossover between subtypes of AN was not frequently observed either, in contradiction with literature (Eckert et al., 1995; Eddy et al., 2002; Strober et al., 1997). However, it has to be mentioned that diagnoses of eating disorders was mainly based on self-reports. It has been argued before that AN patients can not be considered recovered as long as psychological features of the disease have not been investigated and/or have not improved (Couturier & Lock 2006; Jarman & Walsh 1999; Pike 1998). Next to BMI and menses, we therefore investigated the respondent s body image, general psychopathology, mood state as well as social and occupational adjustment five years after admission. We observed that cognitions and psychopathology were still present in the majority of respondents, similar to earlier reports (Clausen 2004; Steinhausen 2002; Windauer et al., 1993). Even weight restored participants showed high levels of anorexic-like cognitions and psychopathology, indicating a continuation of their illness at the psychological level even though they had recovered at the physical level. For instance, body attitude was still severely disturbed in W(C)R respondents, implying (life-)long, perturbations in body appreciation and body familiarity. In fact, BAT levels of W(C)R participants were higher than from healthy controls and not different from the AN-R norm group (Probst 1995). Next to persistent elevated BAT values, the majority of the W(C)R respondents commented that they still regularly restricted their food intake. This further implies that preoccupations with eating behaviour remain present in W(C)R respondents. Probst et al. showed before that BAT values improved significantly in the majority of ED participants one year after admission to their specific eating disorder unit (Probst et al., 1999). He showed that participants who rated themselves as improved/recovered had significantly lower BAT values than participants who rated themselves as unimproved (Probst et al., 1999), which was also found in the present study. As both the Belgium and Dutch units treat severely ill patients, differences in changes in BAT levels might be explained by differences in treatment programmes, although it should be noted that timeframes of the two studies were different (1 year vs. 5 year after admission) and unfortunately our sample lacks a BAT value at admission or at T1. 164

17 Outcome of Anorexia Nervosa: results of a 5 year follow-up study Results from the SCL-90 scale showed that current psychopathological complaints include mainly anxious, depressive and somatic complaints, but also insufficiency of thinking and acting, and suspicion and interpersonal sensitivity. Even W(C)R respondents reported high levels of psychopathology. This might be explained by the fact that restoration of body weight and hormonal signalling in W(C)R respondents reduces alexithymia, and patients thus experience more positive feelings as well as negative feelings than during the period of starvation. The type of psychological complaints corresponded with (self-reported) comorbidities. It has been shown before that a high percentage of AN patients suffers from psychiatric disorder(s) after recovery of body weight (Steinhausen 2002). Affective disorders, anxiety disorders (including OCD), personality disorders and substance abuse are among the most observed (Halmi et al., 1991; Herpertz-Dahlmann et al., 2001; Lowe et al., 2001). Our results are therefore in line with findings in the literature, although substance abuse was not reported by the respondents. The participants mood changed over time, but interestingly did not always improve five years after admission. Participants reported to be more tired and tense at T5 than previously, which seems paradoxical but might be explained by alexithymia in the (acute) illness state. NR participants performed worse on mood profiles than W(C)R participants. Despite the frequent presence of psychological complaints, the majority of the respondents rated themselves as happy and reintegrated in social and occupational life, and contrary to our expectations and to previous reports (Herpertz-Dahlmann et al., 2001), differences in outcome (W(C)R, NR) were not reflected in differences in social or occupational adjustment. Interestingly, we found no differences between the 31 WCR and 6 WR participants, except for the presence of menses. Thus although WR participants are not yet somatically recovered, they show similar body psychopathology, body attitude, and mood as participants who recovered both in body weight and menses. We have demonstrated before that WR participants significantly differ from WCR participants in leptin levels and body composition, which validates the use of three outcome categories. Hence, although the small number of WR participants has to be kept in mind, this study also suggests that menses per se is not needed for changes in psychological state. 165

18 Chapter 8 As expected, a relapse experienced before T5 appeared to be correlated with the outcome at T5. We found however no other predictors of recovery in the total participant group. In contrast to Strober, hyperactivity was not identified as a predictor of outcome (Strober et al., 1997). However, W(C)R participants who changed into NR at T5 had higher activity levels at T1, as measured by trained nurses. (We have previously showed that trained nurses can reliably rate patients activity levels and that their ratings correlate with activity watch output (Van Elburg AA 2007)). Despite our earlier finding of an association between patients age at admission and outcome at T1, we found no relationship between age at admission and outcome at T5, suggesting that the outcome at T1 was not stable (van Elburg et al., 2007). Several possible predictors of outcome of AN have been identified before, e.g. duration of the illness, age at onset, body weight at admission, minimum weight, extent and intensity of the initial symptoms and disabilities, (duration of) hospitalization, vomiting, poor social relating or problems in the family, and personality disturbances (Ben Tovim et al., 2001; Fichter et al., 2006; Herpertz-Dahlmann et al., 2001; Lowe et al., 2001; Ratnasuriya et al., 1991; Smith et al., 1993; Steinhausen 1997; Steinhausen 2002; Strober et al., 1997; Walford & McCune 1991). Inconsistency is, however, frequently noted, which might be due to differences between the studied patient samples as well as by small sample sizes. Unfortunately our study did not contribute in clarifying this issue as no predictors of outcome were found, however, specific phenotypical information is still very valuable for refining treatment strategies for evidence-based medicine. Therefore further studies on (phenotypical) predictors of treatment outcome using standardized protocols, objective markers, fixed follow-up times and large patient samples should be encouraged. Recently several reports have been published on the putative traits of AN. For instance, recovered AN patients still show high levels of harm avoidance, impaired set-shifting and increased striatal Dopamine receptor 2/3 binding (Frank et al., 2005; Holliday et al., 2005; Klump et al., 2004) The presence of persistent adaptations in anxiety and depression at T5 as found in our study might not only indicate continuing psychopathology, but might also indicate putative traits underlying the development of AN. This study has several pitfalls. The interval between admission and follow-up varied between participants, with an average of 4.8 years, ranging from 3.4 to 6.1 years. However, results appeared not related to the exact time of followup (data not shown). Another pitfall 166

19 Outcome of Anorexia Nervosa: results of a 5 year follow-up study is that the data collected was mainly based on self-reports. Several studies (Cooper et al., 1989; Fairburn & Beglin 1990) have shown that self-report measures for eliciting and defining eating disorder symptoms are prone to bias and inaccuracies and therefore inferior to clinical interviews. Rohde and colleagues, however, showed that consistency varies according to the psychological problem that is being assessed (Rohde et al., 1997). For instance, agreement was excellent for self- reporting of anxiety disorders and very good for depressive disorders. The data we gathered on eating disorder symptoms may be debatable. Although some items of eating behaviour were questioned, a standard eating disorder inventory (EDI-2) or psychiatric interview was not performed. We did, however, investigate anorexic cognitions by using the body attitude questionnaire and showed that anorexic cognitions are still present in the majority of the participants. However, the information on body weight and menses was always double checked or evaluated by an expert (AvE). The expert continues to have, even 5 years after admission, regular contact with the patients and/or their parents, and is often aware of the current illness state. Another limitation of this study is the phenomenon of missing data. For instance, some respondents reported that they stopped weighing themselves, resulting in missing values on the parameter body weight. Another important consideration is that our results may not be extrapolatable to all AN patients. The follow-up study was performed with AN patients that were referred to our tertiary specialized eating disorder units, most often after unsuccessful treatment elsewhere, and includes only patients who also joined the longitudinal follow-up study. Likewise, the majority of the participants also continued treatment elsewhere following treatment in our centers because the center s policy is to refer patients to regional treatment facilities once sufficient improvement has been achieved. Finally, recovery numbers stand or fall with the criteria used. We determined recovery using two parameters, namely BMI and menstrual cycle, and classified participants as weight and cycle recovered (WCR), weight recovered (WR) or not recovered (NR). A minimum BMI of 18.5 was considered a recovered BMI and presence of regular menses was considered a recovery of menstrual cycle, based upon growth tables of healthy Dutch adolescents/adults (TNO). These criteria were also used in the first year of this longitudinal study (van Elburg et 167

20 Chapter 8 al., 2007). Off course, different recovery numbers would have been obtained when different criteria would have been used. Even in this relatively young sample AN has dominated many years of our study participants lives. The impact of the illness on the patients as well as their family and finances is substantial. With this study we attempted to describe the long-term outcome of their treatment. Five years after admission, the majority of participants had a good somatic outcome, but many of them still suffered severely from anorexic cognitions, anxiety, and depression. Despite persistent psychopathology, the majority of the participants reported that they are enjoying life and reintegrated in social and occupational aspects. We unfortunately did not discover any predictors of recovery but did ascertain that one year treatment outcome was unstable. Therefore we recommend further studies with repeated measurements of AN outcome which may contribute to development of evidence-based treatments in future. 168

21 Outcome of Anorexia Nervosa: results of a 5 year follow-up study References 1. Arrindell WA&EJHM. Symptoms Checklist; handleiding bij een multidimensionale psychopathologieindicator Amsterdam, Ettema & Harcourt Assesment B.V. Ref Type: Generic 2. Ben Tovim DI, Walker K, Gilchrist P, Freeman R, Kalucy R, Esterman A (2001). Outcome in patients with eating disorders: a 5-year study. Lancet 357: Birmingham CL, Su J, Hlynsky JA, Goldner EM, Gao M (2005). The mortality rate from anorexia nervosa. Int J Eat Disord 38: Bulik CM, Sullivan PF, Fear J, Pickering A (1997). Predictors of the development of bulimia nervosa in women with anorexia nervosa. J Nerv Ment Dis 185: Clausen L (2004). Time course of symptom remission in eating disorders. Int J Eat Disord 36: Cooper Z, Cooper PJ, Fairburn CG (1989). The validity of the eating disorder examination and its subscales. Br J Psychiatry 154: Couturier J, Lock J (2006). What is recovery in adolescent anorexia nervosa? Int J Eat Disord 39: Eckert ED, Halmi KA, Marchi P, Grove W, Crosby R (1995). Ten-year follow-up of anorexia nervosa: clinical course and outcome. Psychol Med 25: Eddy KT, Keel PK, Dorer DJ, Delinsky SS, Franko DL, Herzog DB (2002). Longitudinal comparison of anorexia nervosa subtypes. Int J Eat Disord 31: Fairburn CG, Beglin SJ (1990). Studies of the epidemiology of bulimia nervosa. Am J Psychiatry 147: Fennig S, Fennig S, Roe D (2002). Physical recovery in anorexia nervosa: is this the sole purpose of a child and adolescent medical-psychiatric unit? Gen Hosp Psychiatry 24: Fichter MM, Quadflieg N, Hedlund S (2006). Twelve-year course and outcome predictors of anorexia nervosa. Int J Eat Disord 39: Frank GK, Bailer UF, Henry SE, Drevets W, Meltzer CC, Price JC, Mathis CA, Wagner A, Hoge J, Ziolko S et al (2005). Increased dopamine D2/D3 receptor binding after recovery from anorexia nervosa measured by positron emission tomography and [11c]raclopride. Biol Psychiatry 58: Halmi KA, Eckert E, Marchi P, Sampugnaro V, Apple R, Cohen J (1991). Comorbidity of psychiatric diagnoses in anorexia nervosa. Arch Gen Psychiatry 48: Herpertz-Dahlmann B, Muller B, Herpertz S, Heussen N, Hebebrand J, Remschmidt H (2001). Prospective 10-year follow-up in adolescent anorexia nervosa--course, outcome, psychiatric comorbidity, and psychosocial adaptation. J Child Psychol Psychiatry 42: Herpertz-Dahlmann BM, Wewetzer C, Schulz E, Remschmidt H (1996). Course and outcome in adolescent anorexia nervosa. Int J Eat Disord 19: Herzog DB, Keller MB, Lavori PW (1988). Outcome in anorexia nervosa and bulimia nervosa. A review of the literature. J Nerv Ment Dis 176: Holliday J, Tchanturia K, Landau S, Collier D, Treasure J (2005). Is impaired set-shifting an endophenotype of anorexia nervosa? Am J Psychiatry 162: Jarman M, Walsh S (1999). Evaluating recovery from anorexia nervosa and bulimia nervosa: integrating lessons learned from research and clinical practice. Clin Psychol Rev 19:

22 Chapter Kaye WH, Bulik CM, Thornton L, Barbarich N, Masters K (2004). Comorbidity of anxiety disorders with anorexia and bulimia nervosa. Am J Psychiatry 161: Keel PK, Dorer DJ, Franko DL, Jackson SC, Herzog DB (2005). Postremission predictors of relapse in women with eating disorders. Am J Psychiatry 162: Klump KL, Strober M, Bulik CM, Thornton L, Johnson C, Devlin B, Fichter MM, Halmi KA, Kaplan AS, Woodside DB et al (2004). Personality characteristics of women before and after recovery from an eating disorder. Psychol Med 34: Kreipe RE, Churchill BH, Strauss J (1989). Long-term outcome of adolescents with anorexia nervosa. Am J Dis Child 143: Little K, Penman E (1989). Measuring subacute mood changes using the profile of mood states and visual analogue scales. Psychopathology 22: Lowe B, Zipfel S, Buchholz C, Dupont Y, Reas DL, Herzog W (2001). Long-term outcome of anorexia nervosa in a prospective 21-year follow-up study. Psychol Med 31: Morgan HG, Russell GF (1975). Value of family background and clinical features as predictors of long-term outcome in anorexia nervosa: four-year follow-up study of 41 patients. Psychol Med 5: Noordenbos G, Seubring A (2006). Criteria for recovery from eating disorders according to patients and therapists. Eat Disord 14: Norring CE, Sohlberg SS (1993). Outcome, recovery, relapse and mortality across six years in patients with clinical eating disorders. Acta Psychiatr Scand 87: Pike KM (1998). Long-term course of anorexia nervosa: response, relapse, remission, and recovery. Clin Psychol Rev 18: Probst M, Vandereycken W, Van Coppenolle H, Pieters G (1999). Body experience in eating disorders before and after treatment: a follow-up study. Eur Psychiatry 14: Probst MVWVCHVJ. Body Attitude Test for patients with an eating disorder: psychometric characteristics of a new questionnaire. Eating disorders: The Journal of Treatment and Prevention. 3, Ratnasuriya RH, Eisler I, Szmukler GI, Russell GF (1991). Anorexia nervosa: outcome and prognostic factors after 20 years. Br J Psychiatry 158: Rohde P, Lewinsohn PM, Seeley JR (1997). Comparability of telephone and face-to-face interviews in assessing axis I and II disorders. Am J Psychiatry 154: Smith C, Feldman SS, Nasserbakht A, Steiner H (1993). Psychological characteristics and DSM-III-R diagnoses at 6-year follow-up of adolescent anorexia nervosa. J Am Acad Child Adolesc Psychiatry 32: Steinhausen HC (1997). Outcome of anorexia nervosa in the younger patient. J Child Psychol Psychiatry 38: Steinhausen HC (2002). The outcome of anorexia nervosa in the 20th century. Am J Psychiatry 159: Steinhausen HC, Glanville K (1983a). A long-term follow-up of adolescent anorexia nervosa. Acta Psychiatr Scand 68: Steinhausen HC, Glanville K (1983b). Follow-up studies of anorexia nervosa: a review of research findings. Psychol Med 13: Steinhausen HC, Seidel R (1993). Outcome in adolescent eating disorders. Int J Eat Disord 14:

23 Outcome of Anorexia Nervosa: results of a 5 year follow-up study 40. Strober M, Freeman R, Morrell W (1997). The long-term course of severe anorexia nervosa in adolescents: survival analysis of recovery, relapse, and outcome predictors over years in a prospective study. Int J Eat Disord 22: Tozzi F, Thornton LM, Klump KL, Fichter MM, Halmi KA, Kaplan AS, Strober M, Woodside DB, Crow S, Mitchell J et al (2005). Symptom fluctuation in eating disorders: correlates of diagnostic crossover. Am J Psychiatry 162: van Buuren S, Fredriks M (2001). Worm plot: a simple diagnostic device for modelling growth reference curves. Stat Med 20: van der Ark LA, Marburger D, Mellenbergh GJ, Vorst HCM, Wald F (1995). De aangepast profile of moods state; handleiding en verantwoording. Nijmegen, Berkhout Nijmegen B.V. 44. Van Elburg AA, Kas MHJ, Eijkemans R, Van Engeland H (2007). The impact of hyperactivity and leptin on recovery from anorexia nervosa. Journal of Neural Transmission (in press). 45. van Elburg AA, Eijkemans MJ, Kas MJ, Themmen AP, de Jong FH, van Engeland H, Fauser BC (2007). Predictors of recovery of ovarian function during weight gain in anorexia nervosa. Fertil Steril in press. 46. Walford G, McCune N (1991). Long-term outcome in early-onset anorexia nervosa. Br J Psychiatry 159: Wentz E, Gillberg C, Gillberg IC, Rastam M (2001). Ten-year follow-up of adolescent-onset anorexia nervosa: psychiatric disorders and overall functioning scales. J Child Psychol Psychiatry 42: Windauer U, Lennerts W, Talbot P, Touyz SW, Beumont PJ (1993). How well are cured anorexia nervosa patients? An investigation of 16 weight-recovered anorexic patients. Br J Psychiatry 163:

24